I build security systems for fintech, digital assets and AI.
My work runs from custody and transaction approvals to incident response, audits and vulnerability research. I design the controls and work with the teams that have to operate them.
No theory for theory’s sake — only measures that work in the field.
Selected work
All projects →AI security · In development
Who gets to decide what an agent can do?
I’m building a control layer between an AI agent and its tools. Permission checks run outside the model; execution records retain the policy decision and the action taken.
Read the architecture note →Fintech · Vault.ist
From approval to signing
I designed Fireblocks controls for transaction approvals, signing quorums, wallet separation and recovery. The work included preserving the destination address between approval and signing.
Role and responsibilities →Open-source research
Finding the failure in the code
Published findings in Gitea, OpenEXR, FreeRDP and other projects, covering authorisation checks, parser arithmetic and file handling. Each disclosure links to an advisory and its credit.
Explore the disclosures →Selected writing
All notes →Background
I lead security and AI work at Medici Expert / AI Lab. Previously at Crypterium, Choise.com and Vault.ist. I support around ten security and advisory projects a year, mostly in fintech and digital assets.
Experience →